[Openexr-user] Re: Survey of OpenEXR usage

[Dan Maas]

In response to Drew's survey question - I am using EXR with the PIZ
codec as an on-disk storage format for intermediate floating-point
elements. I was using floating-point TIFF previously, but EXR's
compression is much better (the truncation to 16 bits does not make a
difference and the PIZ codec does a great job from there). I use a
custom conversion directly from RGB float to 8- or 10-bit Y'CbCr for
video work, and plan on using plain-vanilla Cineon for upcoming film
work.

I am actually using a custom re-implementation that handles a
compatible subset of the EXR format. (the ILM code is very cool but we
have some runtime constraints on the C++ dialect, e.g. avoiding STL
usage).  At some point I was thinking about releasing this code, if
only as a check for file format compatibility. (please do not bug me
about it though, it's waaaaay down my priority list).

I would like to echo Thad Beier's earlier post - I am a little leery
of requiring some kind of interpretation data to understand the
numbers contained within an EXR file. In my personal opinion, an image
file format is a container for numbers, nothing more, nothing less.
The computer graphics industry is blessed with the tradition that we
are always aware of what the actual numbers are.

I would love to see a variant of Cineon with lossless compression
(something like the new Pxr24 codec would probably be very effective)
but that is a separate issue.

Regards,
Dan